The neuroethics of memory

Inwhat sense would I still be the same person if my memories werereplaced by false ones?. Dramatic poten-tial alterations include the deliberate deletion of memories, or theinsertion of

5 The neuroethics of memory

One scenario which simultaneously fascinates and horrifies manypeople is the prospect that our memories could be altered by others.The number of films depicting this kind of scenario bears witness toits fascination; think of Total Recall, Eternal Sunshine of the Spot-less Mind or Dark City The prospect of losing our memories, orhaving them replaced with false recollections, exerts such power over

us because we all recognize, more or less clearly, that our memoriesare, in some sense, us: our very identities (in one sense of thatmultiply ambiguous term) are constituted by our past experiences

behavior, thoughts and desires

The so-called memory criterion of personal identity was ginally proposed by John Locke, the great seventeenth-century Eng-lish philosopher Locke argued that a person at time t was the sameperson as an individual at some earlier time if at t they are able toremember experiences of that earlier individual Locke’s criterioncame under attack almost immediately, and with good reason: phi-losophers like Thomas Reid pointed out that the memory criterionwas circular Memory presupposes personal identity, and thereforecannot constitute it I can only remember things that actually hap-pened to me; that’s part of the very definition of memory (if I seem toremember being abducted by aliens, but I was never in fact abducted

ori-by aliens, I don’t actually remember being abducted ori-by aliens;

‘‘remember’’ is a success word and is only appropriately applied whenthe event actually happened, and the recollection is appropriatelycaused by the event) Nevertheless, Locke was clearly onto someimportant aspect of identity He was wrong in thinking that memory

provides a criterion of persistence of identity across time, but it doesconstitute our identity in a different sense.

Marya Schechtman usefully distinguishes between two senses

of personal identity The traditional debate in philosophy, the one towhich Locke took his memory criterion to be a contribution, seeksanswers to what Schechtman calls the reidentification question: thequestion of whether an individual at t is the same person as anindividual at another time There are circumstances in which thereidentification question actually matters (for instance, we might beconcerned with questions about when individuals come into and goout of existence, because the answers seem to bear on issues like themoral permissibility of abortion and the moral significance of sus-taining the life of individuals in persistent vegetative states) But ineveryday life we are usually far more concerned with Schectman’ssecond question, the characterization question: the question ofwhich mental states and attitudes, as well as the actions caused bysuch states, belong to a person When we talk about someone’sidentity, it is generally this sense of identity we have in mind Think

of the phenomenon of the ‘‘identity crisis’’: someone undergoing anidentity crisis does not wonder whether they are now the same per-son (in the reidentification sense) as another past individual; theywonder whether their values and projects are the kinds of things theycan authentically identify with

In this sense of the word, our identities are very importantlyconstituted by our memories At least, they are constituted by ourbeliefs, plans, policies and values, and these things exist across time.What really matters to me is not just a matter of what I think matters

to me now; it is revealed in my behavior over the long-term This is not

to say that I can’t change my mind – conversions on the road toDamascus really happen, after all But a genuine conversion mustitself be ratified by a long-term change in behavior; a short term con-version is merely an aberration Our identities, in this sense, are dia-chronic entities: I am the sum of my plans and policies; I work towards

a goal and I understand myself in terms of my background – where I’m

coming from, as we say, is where I come from (my religion, my munity, my language group and ethnicity, my family) Memory links

com-my past to com-my future self, and makes me the person I am

Hence, I suggest, our horrified fascination with the idea oflosing our memories Would I survive if my memories evaporated? Inwhat sense would I still be the same person if my memories werereplaced by false ones? Hence the fascination not only with falsememories, but also with amnesia: not just films like Eternal Sun-shine of the Spotless Mind, in which the protagonists deliberatelyerase some of their own memories, but also films like Memento and

50 First Dates, in which characters struggle to cope with catastrophicmemory loss We also see the same horrified fascination in ourresponses to dementia, which we see, rightly or wrongly, as thegradual unravelling of the person themselves

Neuroscientific knowledge and the technologies it mightspawn are relevant to our memory-constituted identities in severalways Most directly, it might give us the means of altering ourmemory systems, in more or less dramatic ways Dramatic (poten-tial) alterations include the deliberate deletion of memories, or theinsertion of false memories; less dramatic alterations include theenhancement of our memories, perhaps beyond their current capa-cities, or the treatment or prevention of memory loss More imme-diately, we may already have the ability to modulate the emotionalsignificance of memories in certain ways; a power that promisesgreat benefits, but which also, used inappropriately, might carry greatrisks Understanding the significance of this power is one of the mostpressing issues in all of neuroethics, simply because the techniquesneeded to put it into practice may already exist We shall explore thisquestion fully later in this chapter; for now, let us turn to thequestion of the insertion or deletion of memories

t o t a l r e c a l l

Is it possible to insert false memories into the mind of a person?Developing the power to alter or insert memories requires a far better

understanding of the precise manner in which memories are encodedthan we currently possess There are two obstacles to our being able

to insert memories, one technical and one conceptual Overcomingthe technical obstacle requires unravelling the mechanisms bywhich memories are stored and retrieved, and then using thisknowledge to develop a technique whereby memories can bemimicked We have made great progress at the first half of this task:

we understand how memories are first stored in the medial temporalsystem, in the form of enhanced connections between neurons, with

a particular pattern of connections constituting a particular memory(though we are far from being able to ‘‘read’’ the memory just byexamining the connections between the neurons) We know thatmemories that persist are transferred out of the medial temporalsystem and distributed across networks in cortical regions (Schacter

1996) Because short-term memories and consolidated memories arestored in different regions of the brain, the ability to recall eventslong past can be preserved even when the ability to lay down newmemories is lost There are cases of patients who, through disease orbrain injury, live in an eternal present, entirely unable to recallevents for more than a few seconds, and who therefore do not knowwhere they are or what they are doing These unfortunates typicallyretain memories of their childhoods; in general, the older the mem-ory, the more resistant it is (a phenomenon known as Ribot’s law).Inserting memories requires not only that we understandmemory storage, how a pattern of neural connections constitutes amemory, but also memory retrieval: how the pattern is reactivated.Here, too, we are making great strides, though perhaps it is fair to saythat we know less about retrieval than about storage There is evi-dence that retrieval is, in part, reconstruction: that what is recalled is

an amalgam of the original event as it occurred, and the retrieval cueswhich prompt recall (Schacter 1996) The memories we recall areinfluenced by the goals we have at the moment of recollection, ourintervening experiences and our reinterpretations Hence, each timethat (ostensibly) the same event is recalled, it will in fact be subtly

(and perhaps not so subtly) different: first, because the retrieval cuewill be different in each case (since the context of retrieval isnecessarily different each time), and therefore the combination ofstored memory and retrieval cue will be unique; and second becausethe stored memory itself, the so-called engram, will have changed bythe very fact of having been recalled As Schacter (1996: 71) puts it,

‘‘we do not shine a spotlight on a stored picture’’ when we recollect;instead, we reconstruct the past event using stored cues.1

Retrieval seems to work through the matching up of a cue to anengram; if there is a sufficient degree of match, the memory isrecalled The process is mediated by a kind of index, which keepstrack of the engrams scattered through cortical regions Inserting amemory would therefore require not merely altering the connectionsbetween neurons in such a manner as to mimic a real engram; it alsorequires that the indexing system be deciphered and mimicked Thetechnical challenge is immense, and may in fact prove insurmoun-table We may never understand memories in sufficient detail toknow what neural connections would be needed to create a falsememory, and even if we one day acquire this knowledge, under-standing what is involved is one thing, being able to recreate itourselves is quite another Insertion of false memories, using directintervention into the brain, is at best a long way off

Suppose that we one day overcome the many obstacles thatcurrently stand between us and the ability to insert false memories

It is probable that even then there would be quite severe limitations

on the content of the false memories that could be inserted, tions that stem from the holism of mental content By the holism ofthe mental, I mean the way in which mental content is usuallyinvolved in manifold meaningful links to related content DanielDennett (1978) has explored the ways in which this holism limits thecontent that could be inscribed directly in the brain, had we thetechnology Suppose that we wanted to implant the false memorythat Patty visited Disneyland with her younger brother when she wasfive, when in fact Patty has no younger brother, and suppose we

limita-know what neural connections must be made in order to create thismemory and how this memory must be indexed to be available forrecall; we go ahead and make the changes required Now suppose weask Patty about the trip: ‘‘Do you recall any holidays with yourbrother when you were a child?’’ What will Patty say? She’ll probably

be confused, saying something like ‘‘I seem to recall a trip to neyland with my brother, but I don’t have a brother.’’ For a propo-sition that must occupy a relatively central position in someone’sweb of beliefs, it’s not enough (apparently) just to wire it in: one mustalso wire in a whole set of related beliefs For Patty to recall a trip toDisneyland with her brother, she must recall that she has a brother,and recalling that will require a large set of related propositions andmemories Recalling the existence of a brother implies recallinginnumerable everyday experiences involving him (sitting down at thebreakfast table together, playing together, and so on) or recalling anexplanation for why the brother was absent from everyday life, andwhere he is now Beliefs do not generally come as isolates in themental economy of subjects Instead, they come in clusters, andthe more central to our identity (in the characterization sense of theword) the larger the cluster Central experiences and relationships donot come all by themselves; instead, they spread their shadow overalmost all of our mental lives

Dis-Patty will expect herself, and will be expected by others, to beable to recall all kinds of information about her brother The isolatedthought, that she went to Disneyland with him when she was five, inthe absence of a whole network of related memories will probablyseem more like a hallucination than a veridical memory or belief.Memories of experiences that are central to our lives imply manyother propositions, beliefs and memories Suppose, then, we attempt

to wire in not just the single memory, that Patty went to Disneylandwith her younger brother when she was five, but enough of thenetwork of beliefs that that memory implies to make it stableenough to be accepted by Patty as veridical We shall probablydiscover that this network needs to be very extensive Each of the

propositions implied by the proposition that Patty went to land with her younger brother when she was five itself impliesfurther propositions: propositions about her brother’s friends, pro-positions about their parents (did they favour her over her brother, orvice versa? Did they worry about him? Was he naughty child?) andthese propositions imply yet further propositions Moreover, some ofthem might conflict with memories and beliefs that Patty alreadypossesses: memories of loneliness, perhaps, or of envy of those withsiblings, memories of being asked whether she had brothers andreplying negatively, and so on Will we delete these memories? If wedon’t we risk the failure of our attempt to wire in the memory: whenPatty realizes that her new memory conflicts with others, she mayrevise one or the other, dismissing it as a dream; the less well-embedded memory (the one with the fewest rational connections toother memories) will probably be the one to go But if we do deletethese memories, we shall also have to delete the network of propo-sitions that these memories imply, and so on for these impliedmemories in turn.

Disney-Inserting a memory or a belief almost certainly does not requirethat we insert every other proposition that it might imply, and eraseevery proposition with which it conflicts Our mental economy isnot so coherent as all that If we were pressed on our beliefs longenough, we could all detect some conflicts within our own web:beliefs about friends that are contradictory (that Janine is selfish; thatJanine has on several occasions gone out of her way to help others atsome cost to herself, or whatever it might be) But the conflicts andincoherencies had better not be too obvious: if they are, either ourweb of beliefs is in danger of unravelling, or our status as (reasonably)rational agents will be under threat One possibility is that Patty willend up looking for all the world like a delusional subject Sufferersfrom the classic delusions – Capgras’ delusion, in which the personbelieves that someone close to them has been replaced by animpostor, Cotard’s delusion, in which they believe they are dead,somotaparaphrenia, in which they deny that a limb is theirs, and so

on – often exhibit the same paradoxical belief structure as mightsomeone who recalls that they went to Disneyland with theiryounger brother, and that they have no younger brother It’s oftenobserved that delusions are in many ways incoherent Someone whobelieves that they are dead retains the usual understanding of death.They know that dead people don’t tell others that they are dead Butthey are apparently untroubled by the discrepancy Similarly, suf-ferers from Capgras’ delusion often fail to exhibit the kinds of emo-tional responses we would expect to the belief that someone close tothem has been replaced by an impostor: they don’t call the police, nor

do they worry where their real husband or wife has gone and howthey’re getting on They are very stubborn in affirming their delu-sional belief, but also sometimes seem to believe the direct opposite.There have been numerous attempts at explaining, or explainingaway, the paradoxical patterns of belief of the deluded (Currie2000;Bayne and Pacherie 2005; Hamilton forthcoming) This debateneed not detain us here All we need do is to note that insertingmemories might undermine the coherence of the subject’s beliefs,and thereby the rationality of the subject, in the same kind of way as

do delusions

It should be noted, however, that the limitations on insertingfalse memories that stem from the holism of the mental will affectsome false memories and beliefs more than others The greater thedegree of conflict between the false memory and existing memoriesand beliefs, the more difficult it will be to insert the memory (andend up with a rational subject) Some memories will cohere quitewell with the subject’s existing beliefs; the holism of the mental willpresent no obstacle to their insertion Anything which could wellhave happened but didn’t (that you had ice-cream cake and a balloon

on your fifth birthday; that you got a parking ticket two years ago –obviously the content of such plausible false memories will vary fromsubject to subject, and from culture to culture) could be insertedwithout this particular problem cropping up The holism of themental is therefore only a limitation on our ability to insert false

memories, not an insurmountable obstacle It is, however, anextremely important limitation, for the following reason: in general,the more important the false memory to be inserted – where impor-tant beliefs are either those central to the agent’s identity, in thecharacterization sense, or those which might be expected to have had

a significant impact on then – the more connections there must bebetween it and subsequent memories and beliefs, and therefore thegreater the difficulties posed by the holism of the mental It will proverelatively difficult to convince someone that they were kidnapped byaliens five years ago (since they should have subsequent memoriesthat depend on that event: memories of telling the police, theirfriends, their doctor; memories of nightmares and fears, and so on).False memories can be important without being deeplyembedded into an agent’s mental economy: some kinds of relativelycommonplace events can be significant Therefore, the holism of themental does not make the implantation of important false memoriesimpossible On the contrary, it is already possible to implant sig-nificant false memories The most promising (if that’s the right word)results in memory insertion today do not involve cutting-edge neu-roscientific techniques They involve much lower-tech techniques ofsuggestion and prompting Elizabeth Loftus has shown that thesetechniques can be quite effective at inducing false memories innormal subjects We are highly suggestible creatures, and suggestible

in surprising ways Loftus discovered, for instance, that recall oftraffic accidents was sensitive to the questions asked of subjects: ifthey were asked how fast the cars were going when they smashedinto each other, they recalled higher speeds than if they were askedhow fast they were going when they hit one another; moreover, theywere more likely falsely to recall seeing broken glass if asked theformer question (Loftus 2003) Hundreds of studies have now beenpublished showing that subjects exposed to false information aboutevents they have personally witnessed will frequently incorporatethat information into their later recollections (Loftus 2003) Mis-leading questions seem to fill gaps in subjects’ recall: the proportion

of ‘‘don’t recall’’ responses drops after subjects are primed withmisleading information, and the false information takes the place ofsuch responses.

Loftus has even been able to create memories out of wholecloth In one famous study, she had family members of subjectsdescribe to her events from the subjects’ childhoods She then retoldthese stories to the subjects; unbeknownst to them, however, sheadded one false recollection (of having been lost in a shopping mall atage five, including specific details about how upset they were andhow they were eventually rescued by an elderly person) Abouttwenty-five percent of subjects falsely recalled the event; manyclaimed to recall additional details (Loftus and Pickrell1995) Laterwork, by Loftus and others, now suggests that the proportion ofpeople who will confabulate false memories in this kind of paradigm

is actually slightly higher: around thirty-one percent Moreover, thefalse memories need not be banal: people may confabulate unusualand traumatic false memories (Loftus 2003) To ensure that thememories are truly false, and not actual events recalled only by thesubject, impossible events are sometimes suggested For instance,subjects were brought to recall meeting Bugs Bunny at Disneyland(Bugs is a Warner Brothers character, and would never be found atDisneyland) Sometimes the false memories are very rich and highlyelaborated; often the subject expresses great confidence in theirveracity

These memory distortions also occur, unfortunately, outsidethe laboratory Gazzaniga (2005) provides a recent and strikingexample In 2002, Washington D.C and neighboring Virigina andMaryland were terrorized by a sniper, who for three weeks targetedrandom individuals, killing ten During these panicked three weeks,several witnesses reported seeing the sniper driving a white truck

In fact, the sniper drove a blue car What happened? First, a witnesswho had seen a white truck near the scene of one of the shootingsfalsely recalled seeing the sniper in the truck The media picked

up on the false recollection, and broadcast descriptions of the truck

The expectation that a white truck was involved then primedwitnesses’ memories, leading them to falsely recollect seeing such atruck (2005: 125).

In fact, our memories are far less reliable then we typicallythink We incorporate false information and suggestions, advertently

or inadvertently given to us by others, into our memories; we createcomposite memories out of similar scenes; we transpose details andeven central incidents from one memory to another, and our expec-tations lead us to recall details that never occurred Skilful manip-ulators can use these facts about us deliberately to distort memories;more frequently clumsy interrogators inadvertently lead others tofalsely recall events that did not take place There is plentiful evi-dence that police interrogation of eyewitnesses sometimes leads tofalse recollections Part of the evidence is circumstantial, and comesfrom studies of convictions later overturned on the basis of DNAevidence In one study of forty such cases, fully ninety percent of theconvictions were based at least in part on eyewitness testimony(Gazzaniga 2005: 131) In some cases, of course, the witness mayhave lied, but often they seem just to have got it wrong

The unreliability of eyewitness testimony, at least as it iscurrently elicited, is extremely significant, given that an estimated

75 000 cases annually are decided on the basis of such testimony inthe United States alone What mechanisms lead people falsely torecall seeing a suspect at the scene of a crime? An important part ofthe explanation lies in the manner in which the source of a recol-lection can be forgotten A witness to a crime may be shown mug-shots before they are asked to pick out the suspect from a line up.They may then confidently identify a suspect, recognizing that theyhave seen him before But they may have seen his picture in the mugshots, rather than seeing him at the scene of the crime Suspects haveeven come to mistake their imaginings about a crime scene forrecollections of it, and have consequently confessed to a crime theydid not commit (Schacter1996) Even memories of traumatic events,which seem seared into our brains, can come to be distorted over

time (though such traumatic memories tend to be resilient in theircentral features) Since memories are so easily contaminated, theyshould only be relied upon, in a legal setting, if they are elicitedsensitively; unfortunately, police are largely ignorant of the need toavoid such contamination and may sometimes inadvertently leadwitnesses to confuse what they saw with what the interrogatorsdescribe or suggest.

Advertent or inadvertent memory distortion also occurs inother contexts Consider the phenomenon of repressed memory andits recovery The repression of memory may indeed be a real phe-nomenon: there is evidence that people sometimes do becomeamnesic for traumatic events (Schacter1996) But there is much lessevidence that they can recover these memories, and the evidence thatexists is equivocal and open to dispute On the other hand, the evi-dence that memories, however sincerely held and detailed, can beentirely false, is overwhelming Importantly, we have no way todistinguish real memories from false: neither the person whose

‘‘memory’’ it is nor observers, no matter how highly trained, canconfidently distinguish true memories from false Confidence,vividness, detail – none of these factors distinguish true memoriesfrom false.2It may be possible to recover veridical memories of pasttraumas, but unless there is strong independent evidence of theveracity of the recovered ‘‘memory,’’ we should regard such mem-ories with suspicion

The fact that recovered ‘‘memories’’ are more likely than notfalse (perhaps far more likely) matters greatly: there have been manycourt cases, criminal and civil, which are mainly or even entirelybased around such recollections Many people have gone to jail,convicted largely, often exclusively, on the basis of recovered mem-ories Many of these convictions are certainly wrongful, in the sensethat the person did not in fact commit the crimes of which he or she

is accused; all of them are unsafe Consider one well-known case,that of Paul Ingram Ingram, a sheriff’s deputy, was accused ofmolesting his daughters, on the basis of their recovered memories

(after a member of the daughters’ church, who claimed to have thegift of prophecy, told one of them that God had told her that thedaughter had been abused) Ingram claimed not to recall any abuse,but at the urging of his pastor and a therapist, Ingram began to

‘‘remember.’’ The girls’ accusations rapidly became more bizarre:they claimed they had been raped and forced to bear children, whowere then sacrificed in a Satanic ritual attended by other towns-people Ingram duly recalled many of these supposed events Hepleaded guilty when the case came to trial, and was sentenced totwenty years imprisonment

During the trial, Ingram was extensively tested by RichardOfshe, a social psychologist Ofshe wanted to know how suggestibleIngram was As Loftus had shown, many of us are vulnerable tohaving false memories implanted in us, but could a memory of sexualabuse be created out of whole cloth? Ofshe fabricated a false memoryfor Ingram: he told him that Ingram’s son had reported that he andIngram’s daughter had been forced to have sex in front of him In fact,

no one had made any such allegation Ingram failed to recall theincident at first, but after praying and meditating, he developed adetailed memory of the event In part on the basis of Ofshe’s obser-vations, Ingram attempted to change his plea to not guilty, but hewas too late: the court rejected his plea (Schacter1996) He servedfourteen years before being released in 2003

The guilty parties, in many cases of recovered memories, arethe therapists and counsellors who encourage their emergence.Sometimes recovered memories might appear spontaneously; moreoften, they are coaxed and cajoled by well-meaning but ignorantadvisers Some psychotherapists use techniques – encouragingpatients to visualize events they cannot recall, or to pretend that theyhappened – which are known to be effective in producing falsememories, or in otherwise bringing people to mistake imaginings forreality (Loftus 1993) Recovered memories may, occasionally, beveridical But they seem far more likely to be false, and they arecertainly never reliable enough to serve as the basis for a criminal

conviction When people believe themselves to recover memories ofabuse, they suffer a great deal of pain: the memories are no lesstraumatic for being false, the sense of betrayal by loved ones nosmaller They suffer and their families, often their entire localcommunity, suffers as well Moreover, as Loftus (1993) points out,the cycle of recovered memory and exposure risks producing othervictims: genuine victims of childhood sexual abuse (almost all ofwhom never forget the abuse) may be disbelieved, tarred with thesame brush as those who recover ‘‘memories’’ of abuse.

Reflecting on the ways in which memory is already subject todistortion and manipulation ought thus to give us pause Though thepower that neuroscience might offer to distort memories raises ser-ious moral and political qualms, no less serious are the problems thatcurrently beset us Induced memory distortions, deliberate or acci-dental, already impose high costs: innocent people convicted ofcrimes they did not commit, families torn apart by accusations ofabuse, investigations going astray because witnesses incorporatefalse information into their recollections We do not seem close toany new neuroscientific technology that might help us avoid theseproblems: the drugs currently in development that promise toenhance memory seem not to protect us against its suggestibility.But there are direct ethical implications of our growing knowledgeabout memory, the way it works and the ways in which it fails Weought to be far less trusting of eyewitness testimony The mere factthat someone sincerely claims to recognize the perpetrator of a crimeshould not be sufficient grounds for conviction, all by itself Onlywhen we can be sure that the memory is uncontaminated – that noone has suggested to the witness that this person is the criminal, thatthere has not been a failure of source memory, so that the persongenuinely recognizes the individual, but is mistaken as to where theyhave previously seen them – should we rely upon such testimony.But avoiding such contamination is near impossible This being thecase, eyewitness testimony must be treated sceptically: only when

it is corroborated (by other witnesses, or, preferably, independent

evidence – DNA, fingerprints, cameras, and so on) should it lead toconviction.

m e m o r y m a n i p u l a t i o n

Suppose that significant memory manipulation, of the kind saged in films like Eternal Sunshine and Total Recall, becomespossible What ethical questions would this raise? I have advocatedthe parity thesis throughout this book But the parity thesis does notcommit us to thinking that all the problems that new technologiesmight present have already been anticipated Far from it: the paritythesis only commits us to saying that the mere fact that a technology

envi-is new and unprecedented, or that it involves direct interventionsinto the brain using new neuroscientific knowledge and techniques,does not give us reason to think that it raises new, or even – neces-sarily – especially great, problems But if the direct manipulations ofmemory envisaged in science fiction films ever do become possible,there is good reason to think that the problems they pose wouldindeed be unprecedented

Memory alteration and erasure could cause harm to the personhim or herself, or to others The harms to self which could poten-tially result would not be unprecedented in form, though they may

be unprecedented in degree What is the nature of these harms?Recall the reasons that memory matters so much for us Memory issignificantly constitutive of our identities, in what Schectman callsthe characterization sense of identity Our memories help us to makesense of our actions and our personalities, by situating them in thecontext of an unfolding narrative Our most important actions gettheir significance from their place in this narrative: we engage inthem in order to further projects which have their origin in the past,and which continue into the future I type this sentence in order tofinish this paragraph, and I aim to finish this paragraph in order tofinish this book; and I want to finish this book in order to what?

We can trace the significance of this work to me ultimately all theway to what Sartre (1956) called my fundamental project, which is

my way of living my life Perhaps I aim (at a level slightly lessfundamental than the one Sartre had in mind) to boost my CV and get

a better job, thereby to increase my income and my personal comfort,

or perhaps I aim at recognition from my peers, or at increasing theamount of knowledge in the world In any event, the meaning of myaction is constituted in very important part by the threads that link

my past to my future

Now, as Schectman recognizes, our personal narratives arenever wholly consistent or coherent The suggestibility of memory,and the way it alters with recall, ensures that each of us is likely tomisremember certain events A certain amount of incoherence neednot matter, nor a certain amount of falsity It is a difficult matter toidentify the point at which incoherence or falsity ceases to beinnocuous Part of the problem stems from the fact that more thanone kind of good depends upon the kinds of narratives we constructfor ourselves, and that these goods can conflict Narratives are routes

to self-knowledge, which is a good that is – arguably – intrinsic, that

is, valuable in its own right, as well as a good that is instrumentallyvaluable inasmuch as it allows us to achieve other goods (sinceknowing one’s own strengths and weaknesses allows one to planfuture actions more effectively) But narratives can also be instru-mentally valuable independently of their truth

An example will make this clearer The American philosopherOwen Flanagan (1996b) has related how one of his own childhoodmemories proved instrumentally valuable to him Flanagan had, hetells us, very few friends as a young child But he did have one closefriend, Billy, with whom he spent many happy hours playing Later

he lost touch with Billy, but the memory of this important friendshipgave him the confidence he needed to approach people and make newfriends Years later he discovered that his Billy memories werealmost entirely false Billy had been the son of one of his father’scolleagues, and had visited the Flanagans just once Owen had indeedplayed with Billy, but they had never been friends The close andlong-lasting friendship was a fabrication, built upon the flimsiest of

foundations – yet the influence of the ‘‘memory’’ had been very real.

It really had contributed significantly to Owen’s later success atmaking friends

Flanagan’s memory was false, and therefore could not tribute to his self-knowledge Yet it was instrumentally valuable,helping him to achieve goods that mattered to him More usually, wecan expect instrumental value and intrinsic value to coincide, for thefollowing reason: generally, being able to achieve the goods thatmatter to us depends on having true beliefs, both about ourselves andabout the world If you want to avoid danger, it helps to be able toaccurately distinguish tigers from trees, and to know how good youare at outrunning the former and climbing the latter If Flanagan had

con-no social skills and con-no capacity to develop them, his false memorieswould not have been instrumentally valuable to him; it is onlybecause – by chance – his false memories were not an inaccurateguide to his capacities that they proved useful

The more central a capacity to the agent’s sense of themselves,and the more central to the projects they undertake, the moreimportant it is for them to have an accurate sense of it Flanaganwent on to become a prominent philosopher of mind; it was thereforeimportant to him to have a good sense whether he was more talentedintellectually or on the sporting field For most of us, the kind ofwork we do is central to our sense of identities, in the characteriza-tion sense Friendship, family and relationships are also central tothis sense of identity, and it will be correspondingly important foreach of us to have true beliefs about them It does not matter verymuch whether each of our memories regarding our interactions withthose closest to us are accurate, so long as our general sense of theshape of the relationship – of its narrative course – is accurate(Schechtman 1996) Reminiscing with family and friends we occa-sionally discover events concerning which we have divergent mem-ories It doesn’t matter much, so long as each of us is correct in ourgeneral sense of the importance we have for each other, the place weeach occupy in others’ lives and their affections

However, with regard to many of our memories, truth or sity does matter This suggests one important reason why erasingcertain memories can constitute a significant harm to oneself Some

fal-of the memories we might be tempted to erase are relativelyinconsequential: all those petty embarrassments and humiliationswhich haunt each of us But some of our memories, including some

of our most painful, are important guides to our abilities and itations They constitute self-knowledge, and this self-knowledge

lim-is – at least – instrumentally valuable to us If our lives are to gowell, we require the ability to learn from our experiences: from theways in which we failed, as well as the ways in which we succeeded.Erase our memories, and we leave ourselves at the mercy ofimpulses of the moment The person who erases their failures fromtheir memory risks being a dramatic illustration of Santayana’sdictum that ‘‘those who cannot remember the past are condemned

to repeat it.’’

Though the harms to which we could subject ourselves as aresult of memory erasure would, plausibly, be more dramatic thananything we can currently produce, they would not be entirelyunprecedented People can already learn to cultivate their memories

of incidents, or take steps to repress them Experiments have shownthat subjects instructed to forget items, such as a list of words pre-sented to them, have some success at the task: though their ability torecognize words as having appeared on the list is unaffected, theirability to retrieve them uncued is reduced (Whetstone and Cross

1998) If it is possible to have some effect on one’s own recall over thespace of a single short experiment, it is surely possible to produce amore dramatic effect over a longer time span Nevertheless, this isunlikely to prove as drastic as the effects of the kinds of memoryerasure envisaged in science fiction

If the harms to self made possible by such technologies could

be unprecedented, at least in degree, the harms to others they couldbring would be quite novel Showing that this might be so could proveimportant, because considerations of autonomy normally trump

considerations of self-harm That is, if my actions will harm no onebut myself, then it would normally be wrong for others to coerce meinto refraining from them I have the right to harm myself But I donot have the right to harm others without their consent If, therefore,memory erasure might harm others, we might permissibly preventits use.

How might memory erasure harm others? In several ways Ournarratives, which form the core of our identities, in the character-ization sense, are not merely personal and private stories Each of us

is perpetually at risk of what psychologists call confabulation:inventing more or less plausible stories that bear little relationship toreality to explain what we do and why Without a public check onwhat happens to us, the risk that we shall slip into unchecked fan-tasy is high Once again, the way memory works is important to thisprocess Source memory, our memory for where we acquired a piece

of information, is dissociable from semantic memory, our memoryfor facts (Schacter 1996): it is this fact that accounts, in part, foreyewitness misidentifications The dissociation of source memoryfrom semantic memory explains how people can perfectly innocentlycome to believe their own fantasies, and this, too, has had tragic real-world consequences It is therefore very important, for our self-knowledge, that we associate with people who can corroborateimportant elements of our life story Now, if I erase my memories ofyou, I risk harming myself But I also risk harming you: I removemyself, permanently and irrevocably, as a reference point againstwhich you can check your self-narrative

Moreover, even if we succeed in retaining accurate memories

of the major events in our lives, we would likely be damaged by ourinability to share them with those who feature centrally in itsunfolding narrative In the past decade, philosophers have devoted agreat deal of effort to understanding the concept of recognition, firstintroduced into philosophy by the great German philosopher Hegel.Hegel saw that it was very important for us social animals that ourworth be recognized by other people But, as he also saw, our sense of

self-worth depends upon recognition from others that we ourselvesthink worthy of recognition in turn; moreover, the recognition must

be freely given if it is to be valuable The absence of this kind ofrecognition is, as more recent philosophers have emphasized, pro-foundly damaging to our identity (once again, in the characterizationsense of identity with which we are here concerned)

Why should recognition be so important to our identity? AsCharles Taylor (1995) points out, this is a consequence of the extent

to which our identity is dialogical: we understand ourselves in termswhich we fashion in dialogue with others Whereas other animalshave (at best) only a very rudimentary culture, which does not sig-nificantly shape them, we humans are essentially cultural animals,and culture is by definition something which exists only inter-subjectively: as a result of the interaction of human beings Ouridentity is profoundly cultural, and it is therefore up for negotiationand renegotiation in the stories we tell one another Of course, as wemature we internalize this story-telling; we begin to construct ournarratives for ourselves But, even in cultures that place a high value

on the autonomy of the individual, the extent to which we can everentirely break out of the dialogical mode is limited We engage inlifelong conversations with others – actual conversations, with thosefrom whom we seek recognition, and internal conversations, withthose who have passed from our lives

Now, as profoundly damaging as a lack of recognition ofsomeone’s worth might be, how much worse is the failure even toacknowledge that they have played a significant role in one’s life atall? We cannot demand recognition of others: if relationships fail, wecannot prevent our lovers from walking out We do not have a right

to their time and affection, nor even to their attention Sometimes,

we must move on But we can reasonably expect that our formerlovers will at least acknowledge that we once had a relationship Wesometimes advise friends to try to forget their mistakes, to put thementirely behind them In the worst cases, we might mean it literally:one might do better to entirely forget an abusive relationship (so long